Face Recognition and Verification are widely used for security and identification in recent years. So using this model, users can recognize a person's face and verify the person from the stored database. This project developed for attendance system using face Recognition.
Using transfer-learning, I have take VGG16 model for extracting the facial features. Which will be used to compare to a person's face and verify by choosing threshold value. code
- I have used OpenCV haarcascade_frontalface_default for getting face location coordinate from the input image.
- Now we have to create a vector of all the faces using the transfer-learning concept, and these face vectors are to be compared with existing face_vectors stored in face_vectors directory.
- For comparison of two vectors, I have used Euclidean distance between both vectors. If this distance is less than some threshold (e.g., epsilon=45), then we say they are the same person.
- For creating dataset run dataset_creater
- Run main.py for face verification .